Maximize K-Clustering Spacing: Difference between revisions

From NovaOrdis Knowledge Base
Jump to navigation Jump to search
 
(12 intermediate revisions by the same user not shown)
Line 5: Line 5:
* [[Clustering_Concepts#Clustering_Algorithms|Clustering Concepts]]
* [[Clustering_Concepts#Clustering_Algorithms|Clustering Concepts]]
=Overview=
=Overview=
This is a greedy algorithm that aims to maximize
This is a greedy algorithm that aims to maximize [[Clustering_Concepts#Spacing_of_a_K-Clustering|spacing]] between any two clusters. By spacing between two clusters we mean the minimum distance between any two separated points, which belong to two distinct clusters:
<font size=-1>
min<sub>separated p, q</sub>d(p,q)
</font>
"Good" clustering means that all of the separated points should be as far apart as possible.


=Algorithm=
=Algorithm=
The conceptual pseudocode to achieve k-clustering with maximum possible spacing is:
<font size=-1>
initially each point is a separate cluster
repeat until only k clusters:
  let p,q = closest pair of separated points <font color=teal># determines the current spacing</font>
  merge the clusters containing p and q into a single cluster
</font>
This algorithm is a [[Clustering_Concepts#Single-Link_Clustering|single-link clustering]] algorithm, in that it refines the clustering by fusing clusters one point at a time. It is very similar in structure to the [[Kruskal's Algorithm|Kruskal's algorithm]]. A practical implementation that sorts the distances in a pre-processing steps, in a similar manner to how the Kruskal's algorithm does it, and then manages clusters with an [[Union-Find#Overview|union-find]] is:
<font size=-1>
initialize union-find U <font color=teal># each point belongs initially to its own separate one-point cluster</font>
sort distances <font color=teal># distances between point are sorted from smallest to largest</font>
for d=(p,q) in sorted_distances:  <font color=teal># process distances from smallest to largest</font>
  if U.find(p) != U.find(q): <font color=teal># points belong to different clusters</font>
    if U.clusterCount() == k:
      <font color=teal># we reached the target cluster count, return the spacing between clusters</font>
      return d
    else:
      U.union(cluster(p), cluster(q)) <font color=teal># merge the clusters</font>
</font>
==Implementation==
{{External|https://github.com/ovidiuf/playground/blob/master/learning/stanford-algorithms-specialization/11-clustering-spacing/src/main/java/playground/stanford/clustering/MaximizeKClusteringSpacing.java}}
=Correctness Proof=
=Correctness Proof=
{{External|https://www.coursera.org/learn/algorithms-greedy/lecture/7lWTf/correctness-of-clustering-algorithm}}

Latest revision as of 04:30, 24 October 2021

External

Internal

Overview

This is a greedy algorithm that aims to maximize spacing between any two clusters. By spacing between two clusters we mean the minimum distance between any two separated points, which belong to two distinct clusters:

minseparated p, qd(p,q)

"Good" clustering means that all of the separated points should be as far apart as possible.

Algorithm

The conceptual pseudocode to achieve k-clustering with maximum possible spacing is:

initially each point is a separate cluster
repeat until only k clusters:
  let p,q = closest pair of separated points # determines the current spacing
  merge the clusters containing p and q into a single cluster

This algorithm is a single-link clustering algorithm, in that it refines the clustering by fusing clusters one point at a time. It is very similar in structure to the Kruskal's algorithm. A practical implementation that sorts the distances in a pre-processing steps, in a similar manner to how the Kruskal's algorithm does it, and then manages clusters with an union-find is:

initialize union-find U # each point belongs initially to its own separate one-point cluster
sort distances # distances between point are sorted from smallest to largest
for d=(p,q) in sorted_distances:  # process distances from smallest to largest
  if U.find(p) != U.find(q): # points belong to different clusters
    if U.clusterCount() == k:
      # we reached the target cluster count, return the spacing between clusters
      return d
    else:
      U.union(cluster(p), cluster(q)) # merge the clusters

Implementation

https://github.com/ovidiuf/playground/blob/master/learning/stanford-algorithms-specialization/11-clustering-spacing/src/main/java/playground/stanford/clustering/MaximizeKClusteringSpacing.java

Correctness Proof

https://www.coursera.org/learn/algorithms-greedy/lecture/7lWTf/correctness-of-clustering-algorithm